Entrepôts, Représentation et Ingénierie des Connaissances
Research Program
The continuous decrease in the price level of storage and transmission of data together with the reliability of information systems shows that the volume of stored numeric data will continue to grow exponentially without us being able to predict the limitations (if they exist). This is something that keeps being confirmed day after day.

Therefore, it has become necessary to propose global methodologies and software platforms that improve, not only the accessibility in terms of speed, but also the understanding of their content. This latter aspect presents the main scientific and technological challenges. Providing a solution to this question is not just a matter of having an "intelligent" computer program. We need to consider this issue from the stage of data acquisition to that of the knowledge extraction, as well as the application of data in a specific environment such as that of marketing or health (diagnostic support in a medical environment).

Aware of the complexity of this challenge, the activities of the laboratory are organised into three complementary dimensions. At the beginning of the chain we find the domain of data warehouse and the related scientific issues, and in the middle of the chain, the mining of these data warehouses. This mining meets new technological and scientific challenges especially when the data are not tabular but heterogeneous (text, image, sound, etc.) having little structure. Finally, at the end of the chain, we find the decision-making aspect which focuses on the implementation of explanatory and predictive models. This chain-relation is not linear, as this description is likely to make us believe. On the contrary, the three aspects are nested and highly dependent. This is the reason why we have organised our team into three axes (ENA-DC, FODA and DECCO) while working on a common objective: accessing the knowledge that is hidden in the data and make it active.